Friday, May 15, 2015

Answer: Bike to work? How many?

Did you get it right? 

Remember that on Wednesday I asked you to guess (without looking!) what fraction of people who commute to work actually ride their bike to work more than half of the time?  

How close were you? 

Not your usual bicycle commuters. The Kaufmann Troupe of "trick cyclists" in the 19th century.  

In particular, we wanted to discover:  

1.  In the whole of the US, what fraction of people ride their bikes to work at least half of the time? 
2.  In your state or province, how do the commute modes (car vs. public transit vs. bike vs. walk) break down?   Can you tell us your local transportation modes use?  

When I started thinking about this, I started trying to imagine what organization would be interested and collect this kind of data.  

First thought: Department of Transportation.  Second thought: Bicyclist lobbying groups.  (I know that bicycling has its own set of lobbyists, so perhaps they'd have some data.) 

My first query was pretty successful: 

     [ ride bike OR bicycle "to work" transportation data  ] 

which took me to the BikeLeague's website and their bike transportation data (from 2013).   

But I also found the data from 2008-20014, which was the source for the Bike League's analysis as well.  That seems like pretty reliable data, what's their number? 

In the report Modes Less Traveled--Bicycling and Walking to Work in the United States: 2008-2012 they show data that suggests that US-wide, the number of bicycle commuters is: 

            0.6% of all commuters.  

(As compared to 2.8% who walk; 4.3% who work from home; and 5.0% who use public transportation.)  

Interestingly, they show this small nationwide increase in bicyclists, but decrease in walkers-to-work:  

Figure 3 from "Modes Less Traveled..." report (2014)

Farther down in that report you can also spot maps that break out the data by state: 

Figure 7 from "Modes Less Traveled..." report (2014)
Unsurprisingly, the southeast doesn't have many bike commuters, while the west coast, Idaho, Montana, New York, and Colorado all have lots of bicyclists.  

As you can see from this chart, the 2008-2012 data comes from the "American Community Survey," so I thought I should spot check the data, and maybe get more up-to-date values.  

The obvious search [ American Community Survey ] brings me to their site, which has a nice Advanced Search feature which allows me to check the 2013 data for California.  From this chart, I'd expect something between 1% and 2%.  Sure enough:  when you check the data table at ACS for California, you find that 1.4% of men and 0.6% of women bicycle to work (for a 1.1% overall rate).  That's lower than I'd like, but it's in the range shown in the above chart.  

So now we know what the Census people have measured, and that the latest ACS data agrees with their numbers as well.  

But we should check these findings.  Question is, could I find some OTHER source of information?  Who else would do this kind of data collection?  

This is an important point because many of the news articles you read all repeat the same data from the same few sources.  We need to get a fact check here, hopefully from a different source with a different survey method.  

By going farther down the results page for the previous query, I found a few reports that use different methods and sources.  

I enjoyed reading the website and looking at their stats from 1997.  (A little old, but useful to compare with the above chart.)  That website points to a study published in the Transportation Research Record (a prestigious journal affiliated with the US National Academy of Sciences).  William E. Moritz "Survey of North American bicycle commuters: design and aggregate results"   Transportation Research Record: Journal of the Transportation Research Board 1578.1 (1997): 91-101.  (PDF)

That paper says, "Estimates of the number of bicycle commuters in the United States range from 0.5 to 2.8 million or approximately 0.4 to 2.3 percent of the total number of commuters. (2,3) These numbers are derived from census data and information contained in the Nationwide Personal Transportation Study (NPTS)."  Those citations (2,3) are from surveys conducted by the Bureau of Transportation in 1990, so they're consistent with the charts above.  

The paper goes on to point out that "... based on the data in this sample, major streets without bicycle facilities have an RDI [Relative Danger Index, where a larger number mean "more dangerous"] of 1.26; minor streets, an RDI of 1.04; streets with bike lanes or bike routes, an RDI of 0.50; bike paths, an RDI of 0.67 and sidewalks, an RDI of 5.30. With the 7.3 million km of bicycle commuting reported, an annual accident rate of 37.1 per million km was calculated..."  [ or 0.0086 crashes / km ]

That's interesting because in yesterday's Bike To Work day, roughly 300 people biked to the Googleplex over a total distance of (this is an estimate) 3000 miles (or 4828 km) and there were 3 reported crashes.  That makes the Google "Bike to Work" crash rate around 0.0006 crashes / km, or roughly 14 times fewer crashes than the national average.  (Of course it helps to have most Googlers be in good shape on high quality roads on nice bikes without rain or snow...)  

Google bikes lined up waiting for riders.

But I digress... 

The point is that we have two different data sources (with different data collection methods) that agree on the bike-commuting rate for 1990.  

These sources also (handily) have breakout data by state.  California has 1.1% bike commuter rate and a 2.7% walkers rate.  (Note that I used to find the walker rate data.  We'll return to FactFinder in another episode.)  

Search Lessons:  Here, the search itself wasn't difficult, other than to use the keywords "bicycle" OR "bike, and "transportation" and "data" to pickup the documents that actually have the data you seek.  

But as you see, it's always a good strategy to second-source your findings.  In this case, I looked for a DIFFERENT source (the Census vs. the Bureau of Transportation) using a somewhat different sampling method.  And as you saw, the results were pretty much exactly the same... which give me confidence that these two methods are measuring the same thing.  

This was fun!  Now I have to ride home.... 

Search on! 

No comments:

Post a Comment